How to speed up huge matrix mldivide?

How to speed up huge matrix mldivide? I was solved this matrix on single core but it took too long. So, i was used distributed array using parallel computing tool box, but i didn't get speed up. Using distributed array took too long than single core. How to speed up solution.

[James Tursa]

How large? Are there any special features of the matrix such as sparsity or symmetry or patterns?

[zawye aung]

n = 5000; M = rand(n); x = ones(n,1);
tic
u = M\x;
clear A b
 disp(['Time to solve on single core = ' num2str(toc)])

% parallel operations in spmd
tic
spmd
m = codistributed(M, codistributor('1d',2)); % by column
y = codistributed(x, codistributor('1d',1)); % by row
v = m\y;
end
disp(['Time to solve on 4 core = ' num2str(toc)])

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Time to solve on single core = 2.4375
Time to solve on 4 core = 4.4646

[zawye aung]

This is a sample coding. And another problem i has. The huge matrix size hurts memory. If run the program using huge matrix with codistributed array in SPMD, the Matlab software is delay because of full of memory.

Answer 1

[Edric Ellis]

distributed arrays on a single machine will generally perform worse than normal arrays - this is because the standard MATLAB mldivide is already intrinsically multithreaded.

If you have access to a capable GPU, and your problem fits in the memory of the GPU, this might be a good option. See this example which compares mldivide performance on the CPU and GPU.

[zawye aung]

Thank you! For your suggestion. I really appreciated that. I had already build MJS cluster, so i executed the mldivide fun, but i didn't get speed up. If i use more core, the program more take time. Any suggestion or guidence.